TDAG is a multi-agent framework that leverages dynamic task decomposition and agent generation to tackle complex problem-solving scenarios. This project encapsulates the implementation details, providing a comprehensive toolset for researchers and developers to explore and extend the capabilities of TDAG.
Follow these instructions to set up the project on your local machine for development and testing purposes.
- Python 3.x
-
Extract the zip file to your desired location.
-
Navigate to the project directory:
cd TDAG -
Install the required dependencies:
pip install -r requirements.txt
No additional configuration is required for the basic setup. For advanced users looking to customize the framework or integrate additional services, please refer to the documentation within the project.
The TDAG framework utilizes a dataset located in ./data/travel for demonstration and testing purposes. This dataset is designed to simulate travel planning scenarios, allowing users to explore the dynamic task decomposition and agent generation capabilities of TDAG.
The framework is divided into two main components: the server setup and the task runner.
python serve.py
### run
python ./run/run_incre.py